Seleziona la regione che meglio si adatta alla tua posizione o alle tue preferenze.
Questa impostazione controlla la lingua dell'interfaccia utente, inclusi i pulsanti, i menu e tutto il testo del sito. Seleziona la tua lingua preferita per la migliore esperienza di navigazione.
Seleziona le lingue per gli annunci di lavoro che desideri vedere. Questa impostazione determina quali annunci di lavoro ti verranno mostrati.
University of Bern – Computer Vision Group
Are you passionate about deep learning, self-supervised learning, generative AI, and pushing the boundaries of artificial intelligence? The Computer Vision Group (CVG) at the University of Bern, Switzerland, invites applications for two PhD positions and two postdoctoral positions across two ambitious research projects.
About the Positions
Successful candidates will conduct original research in one of the following areas:
Project 1: Self-Supervised and Meta-Learning for Rapid Adaptation
2 PhD positions and 1 postdoctoral position
This project investigates how the structure of pretraining data and the choice of learning objective shape the prior knowledge acquired by deep-learning models, and how these priors can be optimized to enable rapid adaptation to novel tasks. The research combines the development of novel datasets, controlled benchmark environments, and methods for evaluating knowledge and problem-solving with the design of new learning algorithms and architectures for efficient, robust, and transferable intelligence. The project is fully funded for four years and will be carried out in close collaboration with Prof. Andrea Vedaldi at the University of Oxford.
Project 2: Collaborative World Models
1 postdoctoral position, joining two PhD researchers who have been already recruited for the project
This project investigates collaborative world-model agents: independently operating visual models that learn to predict how environments evolve, retain distinct perspectives and memories, and communicate to solve problems together. The central question is what these agents should represent and exchange, and how they can align predictions, negotiate plans, and coordinate actions while remaining autonomous and specialized. The research combines generative video modeling, self-supervised learning, memory, test-time adaptation, and decision- making, with potential applications in embodied AI and robotics. The successful candidate will join an existing team working on controllable, memory-augmented, and task-solving world models.
General Information
Your Profile
We are looking for curious and ambitious researchers who are eager to work on high-impact questions in artificial intelligence.
For PhD applicants:
For postdoctoral applicants:
All applicants should have:
Experience relevant to at least one of the projects, including world models, video generation, generative modeling, self-supervised learning, meta-learning, reinforcement learning, multi-agent systems, or embodied AI, is an advantage. Applicants are not expected to have expertise in all these areas.
What We Offer
Interested?
Please submit your application through the CVG submission portal.
In your application, please indicate whether you are applying for a PhD or postdoctoral position and specify the project that interests you. Applicants may express interest in both projects where appropriate.
Applications submitted directly by email will not be considered.
The University which dates back to the XVIth century, has a comprehensive offering including 8 faculties, some 160 institutes and 8 Graduate Schools.
Visita la pagina del datore di lavoro